«

Dave Martindale wrote:

> Ilya Zakharevich
writes:
>>So this price/performance consideration
>>should move in the direction of smaller pixels. If/when the price of
>>extra processing power is close to 0, even small benefits from the
>>added information will make smaller pixels preferable.
>
>
> Again, you're ignoring the fact that smaller pixels are technically
> inferior to larger pixels in sensitivity and noise. Even if computation
> and memory become effectively free, you've still got to deal with this.
>
> Dave

Dave,
A slight correction. Small versus large pixels sizes:
the sensitivity (quantum efficiency) is the same for small
versus large pixels when made from the same technology
(e.g. front side illuminated CCDs). Large pixels collect
more photons simply because they have a larger area.
Like a large bucket set out in the rain will collect more
rain drops than a small bucket.

Roger

rec.photo.digital writes:

>> Again, you're ignoring the fact that smaller pixels are technically
>> inferior to larger pixels in sensitivity and noise. Even if computation
>> and memory become effectively free, you've still got to deal with this.

>Dave,
>A slight correction. Small versus large pixels sizes:
>the sensitivity (quantum efficiency) is the same for small
>versus large pixels when made from the same technology
>(e.g. front side illuminated CCDs). Large pixels collect
>more photons simply because they have a larger area.
>Like a large bucket set out in the rain will collect more
>rain drops than a small bucket.

Yeah, I was being sloppy. What I was really talking about is
"sensitivity at constant S/N", not "sensitivity at X% of full well".
Sensors with larger pixels can be used under consitions of lower
exposure and still obtain a usable image.

There ought to be two proper, separate, clear names for these two
concepts.

Dave

Dave Martindale wrote:

> rec.photo.digital writes:
>
>
>>>Again, you're ignoring the fact that smaller pixels are technically
>>>inferior to larger pixels in sensitivity and noise. Even if computation
>>>and memory become effectively free, you've still got to deal with this.
>
>
>>Dave,
>>A slight correction. Small versus large pixels sizes:
>>the sensitivity (quantum efficiency) is the same for small
>>versus large pixels when made from the same technology
>>(e.g. front side illuminated CCDs). Large pixels collect
>>more photons simply because they have a larger area.
>>Like a large bucket set out in the rain will collect more
>>rain drops than a small bucket.
>
>
> Yeah, I was being sloppy. What I was really talking about is
> "sensitivity at constant S/N", not "sensitivity at X% of full well".
> Sensors with larger pixels can be used under consitions of lower
> exposure and still obtain a usable image.
>
> There ought to be two proper, separate, clear names for these two
> concepts.
>
> Dave

Yes, I agree.

Roger